Multiobjective Optimization of a Supersonic-Inlet Bypass- Duct Splitter via Surrogate Modeling

摘要

The aerodynamic performance of an axisv mmetric supersonic engine inlet is optimized via Kriging surrogate models by changing the splitter geometry that separates the engine and bypass duct regions. The splitter length and parameterized leading-edge shape were the design variables, while the total pressure recovery at the Aerodynamic Interface Plane (AIP) and the peak radial distortion intensity were the two objectives for the optimization problem. Due to the discontinuous nature of using a single function for peak radial distortion at the AIP, this objective was addressed using the The WIND-US computational fluid dynamics (CFD) solver generated 'actual' flow solutions. The Kriging model construction then used these solutions to build the surrogates. The resulting approximate multiobjective, multi-point optimization problem was solved at cruise and takeoff conditions.